Method, system and equipment for monitoring stress and damage of pavement structure and storage medium

Abstract

The invention belongs to the technical field of road engineering and structure health monitoring, and particularly relates to a method, a system and equipment for monitoring stress and damage of a pavement structure and a storage medium. The method comprises the following steps: acquiring space-time reference data, dynamic state data and a vibration acceleration signal of a monitored vehicle; based on the space-time reference data and the dynamic state data, obtaining a dynamic load of each wheel of the monitored vehicle on the road surface contact point; acquiring target data based on the dynamic load and the vibration acceleration signal; updating the pavement structure digital twin model based on target data; acquiring historical multi-source data of a monitored vehicle and inputting the historical multi-source data into the current newest digital twin model of the pavement structure to acquire the accumulated damage degree of the current pavement structure in each grid unit so as to predict the remaining service life of the pavement structure. The invention aims to overcome the problems of monitoring system splitting, limited model precision, insufficient early warning capability and the like in the existing road surface monitoring technology.

Description

Technical Field

[0001] This invention belongs to the field of road engineering and structural health monitoring technology, specifically relating to a method, system, equipment and storage medium for monitoring road surface structural stress and damage, which is particularly suitable for road maintenance, asset management and digital operation and maintenance of intelligent transportation infrastructure. Background Technology

[0002] The health of road infrastructure directly affects traffic safety and transportation efficiency. Traditional pavement health monitoring methods, such as manual inspection, core sampling, and falling weight deflectometer (FWD) testing, generally suffer from drawbacks such as high subjectivity, low efficiency, destructiveness, and difficulty in achieving continuous, real-time monitoring. These methods typically only detect developed surface defects and struggle to identify potential problems such as early micro-damage and stress concentration within the pavement structure in a timely manner.

[0003] With the development of sensor technology, wireless communication, and data processing capabilities, using vehicle or embedded sensors to monitor road surface structural health has gradually become a research hotspot. Existing technologies can be broadly categorized as follows: (1) Wide-area macroscopic assessment based on vehicle-mounted mobile sensing: This type of technology uses vehicles as mobile platforms and collects vehicle vibration response and road surface images through sensors such as accelerometers, GPS, and cameras (e.g., Chinese patent CN109870456A). Its advantages are high detection efficiency and wide coverage, but it is mainly used to evaluate macroscopic characterization indicators such as the International Roughness Index (IRI) and Road Condition Index (PCI). This method is essentially a "reasoning from effect to cause", inferring the road surface condition from the external manifestation of structural response, but it cannot directly and accurately depict the true stress distribution and micro-damage evolution mechanism inside the road surface structure under vehicle load.

[0004] (2) Fixed-point microscopic monitoring based on embedded sensors: This type of technology directly embeds sensors such as strain gauges inside the road structure to obtain stress-strain data at specific locations (e.g., international patent WO2012012903A1). Its advantage lies in its ability to directly measure the internal response of the structure, but it has significant disadvantages such as high deployment costs, limited monitoring range ("point" monitoring), and data that is usually disconnected from dynamic traffic loads and environmental factors. This makes it difficult to extend the monitoring results to the entire road network and to form a closed-loop analysis linked to vehicle loads.

[0005] In recent years, although some technical solutions have attempted to integrate the above two paths through the Internet of Things (IoT), existing integration remains at the "data aggregation" level, that is, uploading macroscopic data from vehicles and microscopic data from fixed points to a cloud platform for correlation. This approach lacks a core, dynamically evolving structural mechanics model as a hub, leading to: 1) Fragmented monitoring process: The three key processes of “vehicle load identification”, “structural stress inversion” and “damage accumulation prediction” are still separate and have failed to form an organic closed-loop monitoring framework. 2) Insufficient integration of models and data: Vehicle-mounted detection solutions rely heavily on empirical indicators or "black box" models, while the data value of embedded solutions has not been fully explored to calibrate and optimize structural models, resulting in insufficient accuracy in stress inversion and damage prediction. 3) Limited early warning capability: The monitoring focus is still on macroscopic damage that has already appeared. There is a lack of effective extraction and analysis capabilities for micro-damage signal fingerprint characteristics caused by stress concentration and characterizing the early stage of structural performance degradation, making it difficult to achieve truly predictive maintenance.

[0006] Therefore, existing technologies have significant gaps in establishing a complete, closed-loop mechanical relationship between "vehicle load → internal structural stress → micro-damage evolution → remaining life prediction". Thus, there is an urgent need in this field for a comprehensive monitoring method and system that can deeply couple vehicle dynamic response with structural models to achieve high-precision stress inversion and utilize micro-damage characteristics for damage accumulation and life prediction. Summary of the Invention

[0007] The present invention aims to overcome at least one of the defects of the prior art and provide a method for monitoring road structure stress and damage, which aims to overcome the problems of fragmented monitoring system, limited model accuracy, and insufficient early warning capability in the existing road monitoring technology.

[0008] The present invention also discloses a system equipped with a method for monitoring stress and damage to road structures.

[0009] The detailed technical solution of this invention is as follows: A method for monitoring stress and damage in pavement structures, the method comprising: S1. Acquire multi-source data of the monitored vehicle, including spatiotemporal reference data, dynamic state data, and vibration acceleration signals; S2. Based on the spatiotemporal reference data and dynamic state data, obtain the dynamic load of each wheel of the monitored vehicle at the road contact point; S3. Obtain the target data, including: Based on the dynamic load, the macroscopic mechanical response of key points inside the road structure corresponding to the vehicle position is retrieved using the current digital twin model of the road structure. Based on the vibration acceleration signal, a neural network model is used to obtain the signal fingerprint associated with a specific type of micro-damage to the road structure and the location information of the signal fingerprint. S4. Update the digital twin model of the road structure based on the target data, including: An optimization objective is constructed based on the macroscopic mechanical response to update the global parameters in the digital twin model of the pavement structure; Based on the signal fingerprint and its location information, the material properties of the finite element mesh at the corresponding location in the digital twin model of the road structure are locally corrected; S5. Acquire historical multi-source data of the monitoring vehicle and input it into the latest digital twin model of the road structure to obtain the cumulative damage degree of the current road structure in each grid cell, so as to predict the remaining service life of the road structure.

[0010] According to a preferred embodiment of the present invention, in step S2, based on the spatiotemporal reference data and dynamic state data, obtaining the dynamic load of each wheel of the monitored vehicle at the road contact point specifically includes: A vehicle multibody dynamics model is pre-calibrated. Based on the spatiotemporal reference data and dynamic state data, the state variables of the vehicle multibody dynamics model are estimated in real time using a Kalman filter, which includes the dynamic compression displacement and compression velocity of the suspension. Based on the state variables, the dynamic load on each wheel of the monitored vehicle at the contact point with the road surface is calculated using a dynamic model.

[0011] According to a preferred embodiment of the present invention, in step S2, the dynamic model includes a nonlinear force-displacement relationship model of the suspension and a dynamic model of the tires, wherein: The expression for the nonlinear force-displacement relationship model of the suspension is:

[0012] In formula (1): This indicates the total support force generated by the suspension; This indicates the dynamic compression displacement of the suspension. Indicates the compression rate of the suspension; This is the linear stiffness coefficient of the suspension. This represents the nonlinear stiffness coefficient of the suspension. is the linear damping coefficient of the suspension. The nonlinear damping coefficient of the suspension; This is a sign function used to ensure that the direction of the damping force is always opposite to the direction of the velocity; The expression for the dynamic model of the tire is:

[0013] In formula (2): This represents the dynamic vertical contact force between the tire and the road surface, i.e., the dynamic load. This refers to the vertical stiffness of the tire. This is the vertical damping coefficient of the tire; This represents the vertical displacement of the unsuspended mass at the center of the wheel; The vertical velocity of the unsuspended mass at the center of the wheel; This indicates the vertical displacement caused by road surface irregularities. This indicates the vertical velocity caused by road surface irregularities. Based on equations (1) and (2), the force balance equations for the unsustainable mass of the wheel are established:

[0014] In formula (3): Indicates the unsustainable mass of the wheel; It is the acceleration due to gravity; This represents the vertical acceleration of the unsustainable mass at the center of the wheel. The dynamic load on each wheel of the vehicle at the contact point with the road surface is calculated based on equations (1) to (3).

[0015] According to a preferred embodiment of the present invention, in step S3, based on the vibration acceleration signal, obtaining a signal fingerprint associated with a specific type of micro-damage to the pavement structure and the location information of the signal fingerprint using a neural network model specifically includes: The vibration acceleration signal is converted into a two-dimensional time-frequency spectrum using a signal transformation method. The time-spectrum map is analyzed using a pre-trained neural network model to automatically identify unique textures or patterns associated with specific micro-damage types and quantify them as signal fingerprints and their corresponding precise geographical locations.

[0016] According to a preferred embodiment of the present invention, in step S4, the objective function of the optimization objective is:

[0017] In equation (4): Let be the objective function. These are the global parameters in the digital twin model of the road structure to be optimized; This represents the total number of data points used for calibration. Indicates the location The parameters used are: The macroscopic mechanical response calculated from the digital twin model of the road surface structure; Indicates the location The actual macroscopic mechanical response of the road surface is indirectly measured or inferred through onboard sensors. For the first The weight of each data point; Represents the square of the L2 norm; For regularization terms, where The regularization coefficient is . This represents the prior value of the global parameter.

[0018] According to a preferred embodiment of the present invention, in step S4, the material properties of the finite element mesh at the corresponding position in the digital twin model of the road structure are locally corrected based on the signal fingerprint and its location information, specifically as follows: Based on the location information of the signal fingerprint, the micro-damage represented by the signal fingerprint is mapped to the corresponding location in the digital twin model of the road structure; The material properties of the finite element mesh at the corresponding location in the digital twin model of the road structure are locally modified.

[0019] According to a preferred embodiment of the present invention, S5 further includes introducing a fatigue damage model, a permanent deformation model, and an environmental impact model for embedding and / or coordinating with the digital twin model of the pavement structure to obtain the cumulative damage degree of the current pavement structure in each grid cell. The fatigue damage model is constructed based on the Miner linear cumulative damage criterion; the permanent deformation model adopts the rutting model in the Tseng-Lytton model or the Vesys model; and the environmental impact model is constructed based on the modulus-temperature prediction equation or the S-shaped function.

[0020] In another aspect of the invention, a system for monitoring pavement structure stress and damage is provided, employing the method for monitoring pavement structure stress and damage as described above, the system comprising: The vehicle-mounted data acquisition module is used to acquire multi-source data of the monitored vehicle in real time during the driving process. The multi-source data includes spatiotemporal reference data, dynamic state data, and vibration acceleration signals. The dynamic load identification module is used to acquire the dynamic load of each wheel of the monitored vehicle at the road surface contact point based on the spatiotemporal reference data and dynamic state data. The parallel data processing module is used to, based on the dynamic load, use the current digital twin model of the road structure to invert the macroscopic mechanical response of key points inside the road structure corresponding to the vehicle position; and, based on the vibration acceleration signal, use a neural network model to obtain signal fingerprints associated with specific micro-damage types of the road structure and the location information of the signal fingerprints. The model adaptive update module is used to construct an optimization objective based on the macroscopic mechanical response to update the global parameters in the digital twin model of the road structure; and to locally correct the material properties of the finite element mesh at the corresponding position in the digital twin model of the road structure based on the signal fingerprint and its location information. The damage accumulation and life prediction module is used to retrieve historical multi-source data of the monitoring vehicle and input it into the latest digital twin model of the road structure to obtain the cumulative damage degree of the current road structure in each grid cell, so as to predict the remaining service life of the road structure.

[0021] In another aspect of the invention, an apparatus is also provided, comprising: At least one processor; and A memory that stores instructions that, when executed by the at least one processor, cause the at least one processor to perform the method described above for monitoring road structure stress and damage.

[0022] In another aspect of the invention, a machine-readable storage medium is also provided, which stores executable instructions that, when executed, cause the machine to perform the method described above for monitoring road structure stress and damage.

[0023] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention provides a method for monitoring stress and damage in pavement structures. By organically integrating precise load identification, stress inversion, micro-damage feature extraction, adaptive model updating, and damage accumulation prediction into a closed-loop system, it achieves the following: (1) Based on closed-loop intelligent monitoring of perception-cognition-prediction, a complete closed loop is constructed from multi-physics field data perception to state cognition through digital twin model, and then to life prediction based on high-fidelity model, which overcomes the fragmentation of existing technologies. (2) Achieve high-fidelity inversion of digital twin models, that is, continuously calibrate digital twin models through dual feedback loops to make them infinitely close to the real state of physical entities, thereby ensuring unprecedented accuracy of stress / strain inversion; (3) Multi-scale damage identification from macroscopic stress to microscopic fingerprints, that is, it can not only analyze macroscopic stress, but also capture early microscopic damage before road structure failure through signal fingerprints, which greatly improves the early warning capability; (4) Prospective life prediction based on dynamic evolution, that is, pavement life prediction is no longer based on a static and unchanging model, but on a dynamic model that is continuously updated and can reflect damage accumulation and state evolution, so that the prediction results are transformed from estimation to accurate extrapolation. (5) It has the advantages of being non-destructive and highly efficient. The entire monitoring process is completed based on vehicles in normal operation, without the need to close traffic, thus achieving efficient, continuous and non-destructive monitoring of the road network. Attached Figure Description

[0024] Figure 1It is a flowchart of the method for monitoring pavement structure stress and damage according to the present invention.

[0025] Figure 2 It is a schematic diagram of the architecture of the system for monitoring pavement structure stress and damage according to the present invention. Detailed implementation manners

[0026] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0027] It should be noted that the following detailed description is exemplary and intended to provide further illustration of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0028] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0029] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0030] To solve the problems existing in the existing pavement monitoring technologies, such as the fragmentation of the monitoring system, the limited accuracy of the model, and the insufficient early warning ability, the present invention provides a non-destructive monitoring method that utilizes on-vehicle multi-source sensor data, accurately identifies vehicle loads, inversely calculates pavement structure stress, and combines an adaptive structure model to predict damage accumulation. The core of this method is to construct a monitoring system that bidirectionally couples and closes the loop feedback between the "Digital Twin model" of the pavement structure and vehicle dynamic response data. This method realizes the continuous and adaptive optimization of the Digital Twin model through two different-scale feedback mechanisms, namely, global parameter calibration based on macroscopic mechanical response and local state correction based on microscopic damage fingerprints, so as to achieve high-precision stress inversion and life prediction.

[0031] The method and system for monitoring pavement structure stress and damage according to the present invention will be further described below in conjunction with specific embodiments.

[0032] Embodiment No. 1 Refer Figure 1 , this embodiment provides a method for monitoring pavement structure stress and damage, and the method includes: S1. Acquire multi-source data of the monitored vehicle, including spatiotemporal reference data, dynamic state data, and vibration acceleration signals.

[0033] In this embodiment, multi-source sensors installed on the monitoring vehicle can be used to collect dynamic data that is strictly synchronized with the vehicle's position in real time during the vehicle's operation. This data includes at least spatiotemporal reference data, dynamic state data, and vibration acceleration signals.

[0034] Specifically, the spatiotemporal reference data can be provided by the fusion of differential GPS (D-GNSS) and inertial measurement unit (IMU) to obtain centimeter-level three-dimensional position, velocity and attitude (roll, pitch, heading) information of the vehicle.

[0035] The dynamic state data can be provided by wheel speed sensors, suspension displacement sensors or strain gauges, and is used to reflect the vehicle's operating state and the dynamic response of the suspension system.

[0036] The vibration acceleration signal can be acquired by a high-frequency triaxial accelerometer installed at key locations on the axle or body to capture vehicle vibration signals. This signal is both the result of the road surface's "excitation" of the vehicle and contains "response" information of the road surface's structural features.

[0037] S2. Based on the spatiotemporal reference data and dynamic state data, obtain the dynamic load on each wheel of the monitored vehicle at the road contact point.

[0038] This step integrates the multi-source data collected in S1 above and accurately inverts the real-time dynamic interaction forces between the vehicle and the road surface through a pre-calibrated vehicle multibody dynamics model. Its key difference from simple static load estimation lies in: S21. State Observation and Fusion. This involves pre-calibrating a vehicle multibody dynamics model and, based on the aforementioned spatiotemporal reference data and dynamic state data, using a Kalman filter to estimate the state variables of the vehicle multibody dynamics model in real time.

[0039] Specifically, extended Kalman filter (EKF) or unscented Kalman filter (UKF) algorithms can be used, taking measurements such as IMU, wheel speed, and suspension displacement as inputs to estimate state variables in the vehicle multibody dynamics model in real time, such as vehicle body buoyancy and suspension compression. It should be noted that this embodiment does not involve adjustments to the Kalman filter in data processing, therefore the process will not be described in detail.

[0040] Furthermore, the vehicle multibody dynamics model described in this embodiment can be a seven-degree-of-freedom (7-DOF) vehicle model. This model can fully describe the main dynamic characteristics of the vehicle during driving, specifically including: The vehicle body has three degrees of freedom: vertical motion (bouncing) along the Z-axis, pitch motion around the Y-axis, and roll motion around the X-axis. Four degrees of freedom for the four wheels: each wheel has independent vertical motion relative to the vehicle body.

[0041] This model treats the vehicle body and wheels as rigid bodies, simulates the suspension system using springs and damping elements, and describes the interaction between the wheels and the road surface using a tire model. This model is well-known to those skilled in the art, and its state equations effectively describe the vehicle's dynamic response to road surface irregularities, thus providing a solid physical foundation for subsequent dynamic load inversion. Of course, depending on the required accuracy, more complex models can be used, such as a fourteen-degree-of-freedom model including a steering system, or a simplified quarter-vehicle model.

[0042] S22. Dynamic load inversion. Based on the aforementioned state variables, the dynamic load at each wheel contact point of the monitored vehicle with the road surface is calculated using a dynamic model.

[0043] Specifically, based on the state variables estimated in S21 above, the three-dimensional dynamic contact force and tire ground pressure distribution at each wheel contact point of the monitored vehicle are accurately calculated through the mechanical relationships of the dynamic model, such as the nonlinear force-displacement relationship of the suspension and the dynamic model of the tire, rather than just the vertical axle load.

[0044] In this embodiment, the nonlinear force-displacement relationship of the suspension is specifically as follows: The suspension forces in a real vehicle are not simply linear spring forces; they typically include nonlinear stiffness characteristics and speed-dependent damping characteristics. Their force-displacement relationship can be approximated by a polynomial function, as shown in the following equation:

[0045] In formula (1): This indicates the total support force generated by the suspension; This represents the dynamic compression displacement of the suspension, a value that can be estimated by the state observer; Indicates the compression rate of the suspension; This is the linear stiffness coefficient of the suspension. This represents the nonlinear stiffness coefficient of the suspension. is the linear damping coefficient of the suspension. The nonlinear damping coefficient of the suspension; The sign function is used to ensure that the direction of the damping force is always opposite to the direction of the velocity; the coefficient ( , , , This information can be obtained in advance through bench testing or actual vehicle calibration of the monitoring vehicle.

[0046] The dynamic model of the tire is specifically as follows: To accurately describe the interaction between the tire and the road surface, this invention preferably employs a lumped parameter tire model, which simplifies the tire as a spring-damped system. For example, a simple linear model or a more complex nonlinear model can be used. A typical linear model expression is as follows:

[0047] In formula (2): This represents the dynamic vertical contact force between the tire and the road surface; This refers to the vertical stiffness of the tire. This is the vertical damping coefficient of the tire; Represents the vertical displacement of the wheel center (non-suspended mass); Represents the vertical velocity at the wheel center (non-suspended mass); This indicates the vertical displacement caused by road surface irregularities. This indicates the vertical velocity caused by road surface irregularities.

[0048] Based on the above model, the calculation process for dynamic loads is as follows: In step S21, the dynamic compression displacement of the suspension has been estimated using a Kalman filter. and its compression speed Based on the geometric relationships of the vehicle's multibody dynamics model, the force balance equations for the unsustainable mass (wheels) can be established. For a single wheel, its vertical force balance equation can be expressed as:

[0049] In formula (3): Indicates the unsustainable mass of the wheel; It is the acceleration due to gravity; This represents the vertical acceleration of the unsuspended mass at the center of the wheel.

[0050] Because the focus is on the contact force between the wheel and the road surface. And suspension support Based on the estimated state variables and It is directly calculated using the suspension model formula (1). Therefore, through the above mechanical equilibrium relationship, the dynamic contact force between the tire and the road surface can be accurately calculated. .

[0051] For three-dimensional dynamic contact force, tire lateral and longitudinal slip models (such as the Pacejka magic formula) can be further considered, and the solution can be obtained by combining vehicle attitude information (roll, pitch). Tire ground pressure distribution can be calculated by using a more refined tire finite element model or pressure distribution model, with the inverted total contact force as input.

[0052] S3. Obtain the target data, including: Based on the dynamic load, the macroscopic mechanical response of key points inside the road structure corresponding to the vehicle position is retrieved using the current digital twin model of the road structure. Based on the vibration acceleration signal, a neural network model is used to obtain the signal fingerprint associated with a specific type of micro-damage to the road structure and the location information of the signal fingerprint.

[0053] In this embodiment, the dynamic load obtained in S2 above is used as input to perform two parallel data processing steps.

[0054] S31. Macroscopic Mechanical Response Inversion. Based on the dynamic load, the macroscopic mechanical response field of key points inside the road structure corresponding to the location of the monitored vehicle is calculated using the current version of the digital twin model of the road structure.

[0055] Specifically, dynamic loads are applied to the current version of the digital twin model of the road structure, such as a high-precision finite element or analytical model. Through transient dynamic analysis, the macroscopic mechanical response field of key points inside the road structure corresponding to the location of the monitored vehicle is calculated, such as the tensile stress / strain at the bottom of the asphalt layer and the compressive stress on the top surface of the base course.

[0056] S32. Microscopic damage signal fingerprint extraction. This involves performing advanced signal processing and pattern recognition on the vibration acceleration signal to extract signal fingerprints characterizing local micro-damage to the pavement structure.

[0057] Specifically, this process includes: S321. Time-frequency analysis: Continuous wavelet transform (CWT) or Hilbert-Huang transform (HHT) is used to convert the one-dimensional vibration acceleration signal into a two-dimensional time-frequency spectrum to identify the instantaneous energy excitation characteristics of high-frequency bands caused by microcracks, microcavities, etc.

[0058] The aforementioned high-frequency instantaneous energy excitation characteristics refer to the specific patterns exhibited by the vibration signal in the time-spectrum diagram when a vehicle passes over minor road surface defects. The vibration signal energy generated by a normal, intact road surface is mainly concentrated in the low-frequency range, such as below 80 Hz, and the energy distribution is relatively stable. When a wheel rolls over a micro-damaged area, a brief, high-frequency impact response is generated. Specifically, these characteristics may manifest in the time-spectrum diagram as follows: Instantaneous vertical "energy spike": At a precise point on the timeline, a bright line or concentrated energy area that extends from low frequency to high frequency appears and is approximately vertical. This usually corresponds to the impact event when a wheel runs over a transverse microcrack or seam. Local “energy islands”: In a certain high-frequency region of the time spectrum, for example, above 80Hz, there is an isolated, short-lived energy accumulation area. This may indicate that the wheel has run over a local loose, small pothole or early pumping area. Energy enhancement of the high-frequency resonance band: When a vehicle passes through a continuous, mesh-like microcrack area, it may excite a certain high-frequency natural frequency of the axle or suspension system, resulting in a horizontal bright band with a long duration appearing in the corresponding frequency band on the time-spectrum diagram.

[0059] Understandably, the term "high-frequency band" is a relative concept, and its specific range depends on the vehicle type, driving speed, and the damage scale of interest. In a preferred embodiment of the invention, considering the response characteristics of micro-damage to conventional asphalt pavement, the "high-frequency band" typically refers to the frequency range of 80 Hz to 500 Hz. Frequency bands below 80 Hz primarily reflect the macroscopic smoothness and long-wavelength irregularities of the pavement, while frequency bands above 500 Hz may be subject to strong interference from vehicle mechanical noise and sensor noise. Therefore, focusing on the 80-500 Hz range as the key analysis frequency band effectively distinguishes the structural response signals caused by pavement micro-damage.

[0060] S322, Feature Learning and Recognition: Using pre-trained deep convolutional neural networks (CNNs) or recurrent neural networks (RNNs), analyze the time-spectrum map to automatically identify unique textures or patterns associated with specific micro-damage types and quantify them into signal fingerprints and their corresponding precise geographical locations.

[0061] It should be noted that this embodiment does not involve adjustments to the transformation process or the neural network structure, therefore the process will not be described in detail.

[0062] S4. Update the digital twin model of the road structure based on the target data, including: An optimization objective is constructed based on the macroscopic mechanical response to update the global parameters in the digital twin model of the pavement structure; Based on the signal fingerprint and its location information, the material properties of the finite element mesh at the corresponding location in the digital twin model of the road structure are locally corrected.

[0063] In this embodiment, step S4 uses two feedback loops to continuously optimize the digital twin model of the road structure using the results of S3.

[0064] The first feedback loop is used for calibrating the global parameters of the model.

[0065] Its goal is to calibrate the macroscopic material parameters of the digital twin model of the road structure, such as the elastic modulus and Poisson's ratio of each structural layer, as well as structural parameters such as layer thickness, so that it can better reflect the overall performance of the road section and its performance in response to environmental changes.

[0066] The implementation method is as follows: construct an optimization problem whose objective function is to minimize the difference between the macroscopic mechanical response calculated by the digital twin model of the road structure and the actual response indirectly measured or inferred in some way.

[0067] The objective function can be defined as the weighted mean squared error between the model's predicted response and the actual measured response, and its expression is as follows:

[0068] In equation (4): Let be the objective function. For example, global parameters in the digital twin model of the pavement structure to be optimized. ,in They represent the first The elastic modulus, Poisson's ratio, and thickness of each structural layer; This is the total number of data points used for calibration (i.e., the locations the vehicle passed through); Indicates the location The parameters used are: The macroscopic mechanical response calculated from the digital twin model of the pavement structure, such as pavement deflection; Indicates the location The actual macroscopic mechanical response of the road surface is indirectly measured or inferred through on-board sensors. For example, the acceleration signal can be converted into displacement through double integration to obtain the measured value of the dynamic deflection of the road surface. For the first The weights of each data point can be set according to the confidence level of the measurement data; This represents the square of the L2 norm, i.e., the sum of squared errors; For regularization terms, where The regularization coefficient is . This represents the prior value of the global parameters. The purpose of introducing this term is to prevent the model from overfitting and to ensure the stability and physical meaning of the optimized solution.

[0069] By solving this optimization problem, a set of optimal global parameters can be found. This enables the digital twin model of the road structure to better reproduce the actual macroscopic mechanical behavior of the road section.

[0070] The second feedback loop is used to "brand" local damage to the pavement structure.

[0071] The goal is to accurately "imprint" the microscopic damage identified in S32 onto the corresponding location in the digital twin model of the pavement structure, thereby achieving refined modeling of local pavement defects.

[0072] The implementation method is as follows: when the above S32 identifies a "signal fingerprint" with a clear geographical location, the system will automatically make local corrections to the material properties of the finite element mesh at the corresponding position in the digital twin model of the road structure.

[0073] For example, for a microcrack fingerprint, the stiffness (elastic modulus) of the corresponding finite element mesh element can be reduced; for a microcavity fingerprint, contact elements can be introduced or the local support stiffness can be reduced.

[0074] Specifically, in finite element analysis (FEA), a "contact element" is a special type of element that does not represent a solid material entity but is used to simulate interactions such as contact, separation, and friction that may occur between two or more surfaces. For example, commercial finite element software such as ANSYS or ABAQUS defines the CONTA series or contact pairs. These elements define the "contact surface" and the "target surface," and specify their normal behavior (e.g., hard contact, soft contact) and tangential behavior (e.g., coefficient of friction).

[0075] In this method, the "micro-voids" or "vacuuming" defects within the pavement structure are essentially caused by the loss of continuous support and bonding between structural layers (e.g., between the asphalt surface layer and the base layer) in localized areas. Therefore, introducing contact units can be used to: Simulating Separation and Closure: Traditional finite element models typically assume that the structural layers are completely bonded. When tiny void fingerprints are identified, contact elements are introduced at the corresponding locations in the digital twin model to simulate the actual behavior of that area under vehicle loads: when the load moves away, there is a gap between the upper and lower layers (separation state); when the load moves closer, the upper layer bends and deforms, re-contacting the lower layer (closing state) and transmitting pressure.

[0076] Achieving nonlinear support: This contact behavior is a highly nonlinear mechanical behavior that cannot be completely replaced by simply reducing stiffness. Introducing contact elements can more realistically simulate this nonlinear support condition of "sometimes supported, sometimes suspended".

[0077] Improving the accuracy of stress calculation: Accurately simulating the contact behavior of void regions is crucial for accurately calculating stress concentrations (such as flexural stress) at the edges of these regions, which are key factors that cause cracks to originate from and propagate from the edges of voids.

[0078] Therefore, introducing contact elements allows for a more realistic reproduction of the nonlinear mechanical behavior of localized pavement voids in the digital twin model, thus providing a foundation for more accurate stress analysis and damage evolution prediction.

[0079] S5. Acquire historical multi-source data of the monitoring vehicle and input it into the latest digital twin model of the road structure to obtain the cumulative damage degree of the current road structure in each grid cell, so as to predict the remaining service life of the road structure.

[0080] Specifically, the stress / strain historical data derived from the above S3 over a long period (such as several months or several years) and subjected to massive vehicle loads are applied to the latest version of the high-fidelity digital twin model of the road structure, which has been continuously updated and optimized by the above S4.

[0081] By combining fatigue damage models of pavement materials (such as the Miner linear cumulative damage criterion), permanent deformation models, and environmental impact models, the cumulative damage degree of the pavement structure at each location is calculated, and its remaining service life (RUL) is predicted.

[0082] The fatigue damage model, permanent deformation model, and environmental impact model can be embedded into or work in conjunction with the digital twin model of the pavement structure. Understandably, the core of the digital twin model of the pavement structure in this solution is a high-fidelity, multi-physics finite element structural model. It describes the pavement geometry, material distribution, and its physical response (such as stress, strain, and displacement) under load and environmental influences.

[0083] The aforementioned fatigue damage model, permanent deformation model, and environmental impact model are mathematical equations based on pavement materials science and mechanics theories. They use the stress / strain history calculated from the digital twin model of the pavement structure as input to calculate the degradation of material properties (damage accumulation) and the accumulation of deformation.

[0084] In short, the digital twin model of the road structure is used to calculate stress, and these three models are used to calculate damage and deformation based on the calculated stress.

[0085] Further, the permanent deformation model: In the field of road engineering, there are various models for predicting the permanent deformation (rutting) of asphalt mixtures. In this embodiment, the rutting model in the Tseng-Lytton model or the Vesys model is preferably adopted. For example, the Tseng-Lytton model relates the permanent strain to the number of load repetitions, stress state, and temperature, and its form is as follows:

[0086] In Equation (5): represents the permanent strain; is the number of load repetitions; and are model parameters related to material properties, stress level, and temperature, and this parameter can be determined by the stress field and environmental data calculated by the digital twin model of the pavement structure.

[0087] The environmental impact model: This model is used to describe the influence of environmental factors (mainly temperature and humidity) on pavement material parameters (such as elastic modulus). For the temperature influence, the Witzcak empirical formula (such as the modulus-temperature prediction equation used in MEPDG) or a simpler S-shaped function (Sigmoidal Function) can be used to describe the relationship between the modulus of asphalt mixtures and temperature. For example, the expression of the S-shaped function is:

[0088] In Equation (6): is the dynamic modulus; is the reduction frequency, which is related to temperature and loading frequency; are all material fitting parameters.

[0089] For the humidity influence, a relationship model between the moisture content and resilient modulus of the base course and subgrade materials can be established. For example, the empirical formula provided in AASHTO 1993 is adopted to relate the change in moisture content to the reduction in modulus.

[0090] Through the above limitations, the cited models are made specific.

[0091] Since the above prediction is based on a dynamic model that truly reflects the decline of pavement macroscopic performance and the distribution of microscopic damage, its results are much more reliable than those obtained based on the static initial model.

[0092] Example 2 Refer Figure 2 , this embodiment provides a system for monitoring pavement structure stress and damage, applying the method for monitoring pavement structure stress and damage as described above. The system includes: The vehicle-mounted data acquisition module 310 is used to acquire multi-source data of the monitored vehicle in real time during the driving process. The multi-source data includes spatiotemporal reference data, dynamic state data and vibration acceleration signals. The dynamic load identification module 321 is used to acquire the dynamic load of each wheel of the monitored vehicle at the road surface contact point based on the spatiotemporal reference data and dynamic state data. Parallel data processing module 322 is used to, based on the dynamic load, invert the macroscopic mechanical response of key points inside the road structure corresponding to the vehicle position using the current digital twin model of the road structure; and, based on the vibration acceleration signal, use a neural network model to obtain the signal fingerprint associated with a specific micro-damage type of the road structure and the location information of the signal fingerprint. The model adaptive update module 332 is used to construct an optimization objective based on the macroscopic mechanical response to update the global parameters in the digital twin model of the road structure; and to locally correct the material properties of the finite element mesh at the corresponding position in the digital twin model of the road structure based on the signal fingerprint and its location information. The damage accumulation and life prediction module 333 is used to retrieve historical multi-source data of the monitoring vehicle and input it into the latest digital twin model of the road structure to obtain the cumulative damage degree of the current road structure in each finite element grid cell, so as to predict the remaining service life of the road structure.

[0093] In the finite element method, a continuous structure is discretized into a large number of interconnected, geometrically simple elements (such as hexahedrons and tetrahedrons), and the collection of these elements constitutes the finite element mesh. The calculation of damage accumulation is performed on each individual mesh element, which yields a refined distribution map of damage in the three-dimensional space of the entire pavement structure, thereby accurately locating the areas with the most severe damage.

[0094] Furthermore, the vehicle-mounted data acquisition module 310 may include a D-GNSS / IMU 311 for acquiring spatiotemporal reference data, a suspension displacement / wheel speed sensor 312 for acquiring dynamic state data, and a high-frequency triaxial accelerometer 313 for acquiring vibration acceleration signals.

[0095] During actual operation, the monitoring vehicle 300 travels at a normal speed on the target road segment. The onboard data acquisition module 310 deployed on the vehicle then begins to operate.

[0096] The D-GNSS / IMU311 acquires and fuses centimeter-level spatiotemporal reference data at a frequency of 200Hz, including vehicle three-dimensional position, velocity, and attitude data.

[0097] The suspension displacement / wheel speed sensor 312 synchronously records the dynamic compression of the suspension and the wheel speed.

[0098] A high-frequency triaxial accelerometer 313 is installed directly above the rear axle of the vehicle to collect triaxial vibration acceleration signals at a frequency of 2048Hz as structural excitation-response data.

[0099] The dynamic load identification module 321 is loaded within the vehicle-mounted edge computing unit 320. The data collected by the vehicle-mounted data acquisition module 310 is sent in real time to the dynamic load identification module 321 within the vehicle-mounted edge computing unit 320.

[0100] The dynamic load identification module 321 runs a state estimation algorithm based on unscented Kalman filtering (UKF), using data from D-GNSS / IMU 311 and suspension displacement / wheel speed sensors 312 as observations to estimate in real time the state variables of a pre-calibrated seven-degree-of-freedom vehicle multibody dynamics model, such as vehicle body heave, pitch, roll, and suspension compression.

[0101] Based on the estimated suspension compression, the three-dimensional dynamic contact force and tire ground pressure distribution at each wheel-road contact point are accurately derived using the nonlinear force-displacement relationship of the suspension and the dynamic model of the tire, thus obtaining dynamic load data. This process has been described in detail in the aforementioned method and will not be repeated here.

[0102] The parallel data processing module 322 is also loaded within the vehicle-mounted edge computing unit 320. In the parallel data processing module 322, the dynamic load output by the dynamic load identification module 321 is used in two parallel processing flows: a) Macroscopic mechanical response inversion: The parallel data processing module 322 downloads the latest version of the road structure digital twin model from the cloud server platform 330. The model is stored in the road digital twin model library 331 and is a refined three-dimensional finite element model.

[0103] The dynamic load identified by the dynamic load identification module 321 is used as a moving load and applied to the model. Through fast transient dynamics solution, the macroscopic mechanical response field inside the pavement structure is calculated, such as the maximum tensile strain at the bottom of the asphalt layer and the compressive stress on the top surface of the base course.

[0104] b) Extraction of "signal fingerprints" of microscopic damage: In parallel, the parallel data processing module 322 processes the vibration acceleration signal from the high-frequency triaxial accelerometer 313.

[0105] First, a high-resolution time spectrum is generated using continuous wavelet transform (CWT).

[0106] A pre-trained convolutional neural network (CNN) model analyzes the temporal spectrogram to identify unique high-frequency energy excitation patterns associated with micro-damage such as lateral microcracks and localized loosening, generating a "signal fingerprint" and attaching a precise D-GNSS geolocation label. This process has been described in detail in the aforementioned method and will not be repeated here.

[0107] The vehicle-mounted edge computing unit 320 uploads the calculated macroscopic response results, the extracted "signal fingerprint" and its location information to the cloud server platform 330 via the 5G network.

[0108] The model adaptive update module 332 is loaded onto the cloud server platform 330, and includes a global parameter calibration submodule 332a and a local damage imprinting submodule 332b. In the cloud, the model adaptive update module 332 performs a dual-feedback loop update, including: Global parameter calibration: This is performed by the global parameter calibration submodule 332a, which retrieves recent temperature and humidity data from historical data and the environmental database 334. By running a Bayesian inference-based algorithm, and using the macroscopic mechanical response (such as the inverted road deflection) uploaded by the vehicle edge computing unit 320 as evidence, it calibrates the global material parameters (such as the temperature correlation modulus of the asphalt layer) in the road digital twin model library 331, so that the model calculation results match the actual response more closely.

[0109] Local damage imprinting: This is performed by the local damage imprinting submodule 332b. When it receives a "signal fingerprint" with location information uploaded by the vehicle-mounted edge computing unit 320, this submodule automatically corrects the local material properties of the finite element mesh at the corresponding geographical location in the road surface digital twin model library 331, i.e., "imprinting". For example, in the area where a microcrack fingerprint is detected, the elastic modulus of the corresponding finite element mesh element is reduced by 20%.

[0110] Specifically, after the onboard system uploads a microcrack "signal fingerprint" with precise geographic coordinates, the cloud server locates one or more finite element mesh elements covering that geographic location within the vast mesh of the digital twin model through coordinate matching. The system then modifies the material properties (such as the elastic modulus, as exemplified here) of these specific located elements, while leaving the element properties of other areas unchanged. This operation achieves a precise "imprint" of localized damage. This process has been detailed in the aforementioned methods and will not be repeated here.

[0111] The damage accumulation and lifespan prediction module 333 is loaded onto the cloud server platform 330. The damage accumulation and lifespan prediction module 333 performs this step periodically (e.g., daily or weekly): Retrieve the latest version of the high-fidelity digital twin model of the road surface structure from the road surface digital twin model library 331, which has been continuously optimized through a dual feedback loop; Long-term stress / strain historical data for this road section is retrieved from historical data and environmental database 334, and is continuously uploaded and accumulated by all monitoring vehicles; By inputting this historical data into a high-fidelity model and applying the standard asphalt pavement fatigue damage equation (Miner criterion) and permanent deformation model, the cumulative damage degree of the pavement in each grid cell is calculated. Finally, based on the current cumulative damage and estimated future traffic volume, the remaining service life (RUL) of each part of the pavement is predicted, and a visualized pavement health "heat map" and maintenance decision recommendations are generated.

[0112] The estimated future traffic volume can be obtained by combining one or more of the following methods: 1. Trend extrapolation based on historical data: The system can use historical data and long-term accumulated traffic flow data in the environmental database 334 (e.g., daily traffic volume, axle load spectrum, etc. collected by monitoring vehicles or other traffic monitoring equipment in this system) to predict the traffic volume growth trend in the future (e.g., the next 1-5 years) through time series analysis methods (such as ARIMA model, exponential smoothing method).

[0113] 2. Integrate with regional development planning: Incorporate external macroeconomic and urban planning data. For example, if a new logistics park, residential area, or commercial center is predicted to be built near the road segment, a step growth factor can be added to the forecasting model to reflect the foreseeable surge in traffic volume brought about by the planning.

[0114] 3. Consider traffic assignment models: In more complex road network-level forecasting, a four-stage model from the field of transportation planning (trip generation, trip distribution, mode classification, and traffic assignment) can be utilized. When the capacity of other roads in the road network changes (such as the construction of new roads or major repairs to existing roads), traffic assignment models can predict how traffic flow will be redistributed, thereby more accurately estimating the future traffic load of target road segments.

[0115] In a simplified embodiment of the present invention, a trend extrapolation method based on historical data is mainly used, combined with a set average annual growth rate, to predict future traffic volume.

[0116] Based on the above, the method of the present invention constructs a dynamically evolving digital twin of the road surface, realizing multi-scale, closed-loop, and intelligent health monitoring from macroscopic performance to microscopic damage.

[0117] Example 3 This embodiment also provides a device, including: At least one processor; and A memory that stores instructions that, when executed by the at least one processor, cause the at least one processor to perform the method described above for monitoring road structure stress and damage.

[0118] In this embodiment, the device may include, but is not limited to: personal computer, server computer, workstation, desktop computer, laptop computer, notebook computer, mobile computing device, smartphone, tablet computer, cellular phone, personal digital assistant (PDA), handheld device, messaging device, wearable computing device, consumer electronic device, etc.

[0119] Example 4 This embodiment also provides a machine-readable storage medium storing executable instructions that, when executed, cause the machine to perform the method described above for monitoring road structure stress and damage.

[0120] Specifically, a system or apparatus equipped with a readable storage medium may be provided, on which software program code implementing the functions of any of the embodiments described above is stored, and the computer or processor of the system or apparatus can read and execute the instructions stored in the readable storage medium.

[0121] In this case, the program code read from the readable medium itself can perform the functions of any of the above embodiments, and therefore the machine-readable code and the readable storage medium storing the machine-readable code constitute a part of this specification.

[0122] Examples of readable storage media include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD-RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer or the cloud via a communication network.

[0123] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0124] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0125] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0126] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0127] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the technical solutions of the present invention, and are not intended to limit the specific implementation of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the claims of the present invention should be included within the protection scope of the claims of the present invention.

Claims

1. A method for monitoring stress and damage in road surface structures, characterized in that, The method includes: S1. Acquire multi-source data of the monitored vehicle, including spatiotemporal reference data, dynamic state data, and vibration acceleration signals; S2. Based on the spatiotemporal reference data and dynamic state data, obtain the dynamic load of each wheel of the monitored vehicle at the road contact point; S3. Obtain the target data, including: Based on the dynamic load, the macroscopic mechanical response of key points inside the road structure corresponding to the vehicle position is retrieved using the current digital twin model of the road structure. Based on the vibration acceleration signal, a neural network model is used to obtain the signal fingerprint associated with a specific type of micro-damage to the road structure and the location information of the signal fingerprint. S4. Update the digital twin model of the road structure based on the target data, including: An optimization objective is constructed based on the macroscopic mechanical response to update the global parameters in the digital twin model of the pavement structure; Based on the signal fingerprint and its location information, the material properties of the finite element mesh at the corresponding location in the digital twin model of the road structure are locally corrected; S5. Acquire historical multi-source data of the monitoring vehicle and input it into the latest digital twin model of the road structure to obtain the cumulative damage degree of the current road structure in each grid cell, so as to predict the remaining service life of the road structure.

2. The method for monitoring road surface stress and damage according to claim 1, characterized in that, In step S2, based on the spatiotemporal reference data and dynamic state data, the dynamic load of each wheel of the monitored vehicle at the road contact point is obtained, specifically including: A vehicle multibody dynamics model is pre-calibrated. Based on the spatiotemporal reference data and dynamic state data, the state variables of the vehicle multibody dynamics model are estimated in real time using a Kalman filter, which includes the dynamic compression displacement and compression velocity of the suspension. Based on the state variables, the dynamic load on each wheel of the monitored vehicle at the contact point with the road surface is calculated using a dynamic model.

3. The method for monitoring road surface structural stress and damage according to claim 2, characterized in that, In S2, the dynamic model includes a nonlinear force-displacement relationship model of the suspension and a dynamic model of the tires, wherein: The expression for the nonlinear force-displacement relationship model of the suspension is: In formula (1): This indicates the total support force generated by the suspension; This indicates the dynamic compression displacement of the suspension. Indicates the compression rate of the suspension; This is the linear stiffness coefficient of the suspension. This represents the nonlinear stiffness coefficient of the suspension. is the linear damping coefficient of the suspension. The nonlinear damping coefficient of the suspension; This is a sign function used to ensure that the direction of the damping force is always opposite to the direction of the velocity; The expression for the dynamic model of the tire is: In formula (2): This represents the dynamic vertical contact force between the tire and the road surface, i.e., the dynamic load. This refers to the vertical stiffness of the tire. This is the vertical damping coefficient of the tire; This represents the vertical displacement of the unsuspended mass at the center of the wheel; The vertical velocity of the unsuspended mass at the center of the wheel; This indicates the vertical displacement caused by road surface irregularities. This indicates the vertical velocity caused by road surface irregularities. Based on equations (1) and (2), the force balance equations for the unsustainable mass of the wheel are established: In formula (3): Indicates the unsustainable mass of the wheel; It is the acceleration due to gravity; This represents the vertical acceleration of the unsustainable mass at the center of the wheel. The dynamic load on each wheel of the vehicle at the contact point with the road surface is calculated based on equations (1) to (3).

4. The method for monitoring road surface structural stress and damage according to claim 1, characterized in that, In step S3, based on the vibration acceleration signal, a neural network model is used to obtain a signal fingerprint associated with a specific type of micro-damage to the pavement structure and the location information of the signal fingerprint. Specifically, this includes: The vibration acceleration signal is converted into a two-dimensional time-frequency spectrum using a signal transformation method. The time-spectrum map is analyzed using a pre-trained neural network model to automatically identify unique textures or patterns associated with specific micro-damage types and quantify them as signal fingerprints and their corresponding precise geographical locations.

5. The method for monitoring road surface stress and damage according to claim 1, characterized in that, In step S4, the objective function of the optimization objective is: In equation (4): Let be the objective function. These are the global parameters in the digital twin model of the road structure to be optimized; This represents the total number of data points used for calibration. Indicates the location The parameters used are: The macroscopic mechanical response calculated from the digital twin model of the road surface structure; Indicates the location The actual macroscopic mechanical response of the road surface is indirectly measured or inferred through onboard sensors. For the first The weight of each data point; Represents the square of the L2 norm; For regularization terms, where The regularization coefficient is . This represents the prior value of the global parameter.

6. The method for monitoring road surface stress and damage according to claim 1, characterized in that, In step S4, the material properties of the finite element mesh at the corresponding location in the digital twin model of the road structure are locally corrected based on the signal fingerprint and its location information. Specifically: Based on the location information of the signal fingerprint, the micro-damage represented by the signal fingerprint is mapped to the corresponding location in the digital twin model of the road structure; The material properties of the finite element mesh at the corresponding location in the digital twin model of the road structure are locally modified.

7. The method for monitoring road surface structural stress and damage according to claim 1, characterized in that, The S5 further includes introducing a fatigue damage model, a permanent deformation model, and an environmental impact model to embed and / or coordinate with the digital twin model of the pavement structure, in order to obtain the cumulative damage degree of the current pavement structure in each grid cell. The fatigue damage model is constructed based on the Miner linear cumulative damage criterion; the permanent deformation model adopts the rutting model in the Tseng-Lytton model or the Vesys model; and the environmental impact model is constructed based on the modulus-temperature prediction equation or the S-shaped function.

8. A system for monitoring stress and damage in pavement structures, employing the method for monitoring stress and damage in pavement structures as described in any one of claims 1 to 7, characterized in that, The system includes: The vehicle-mounted data acquisition module is used to acquire multi-source data of the monitored vehicle in real time during the driving process. The multi-source data includes spatiotemporal reference data, dynamic state data, and vibration acceleration signals. The dynamic load identification module is used to acquire the dynamic load of each wheel of the monitored vehicle at the road surface contact point based on the spatiotemporal reference data and dynamic state data. The parallel data processing module is used to, based on the dynamic load, use the current digital twin model of the road structure to invert the macroscopic mechanical response of key points inside the road structure corresponding to the vehicle position; and, based on the vibration acceleration signal, use a neural network model to obtain signal fingerprints associated with specific micro-damage types of the road structure and the location information of the signal fingerprints. The model adaptive update module is used to construct an optimization objective based on the macroscopic mechanical response to update the global parameters in the digital twin model of the road structure; and to locally correct the material properties of the finite element mesh at the corresponding position in the digital twin model of the road structure based on the signal fingerprint and its location information. The damage accumulation and life prediction module is used to retrieve historical multi-source data of the monitoring vehicle and input it into the latest digital twin model of the road structure to obtain the cumulative damage degree of the current road structure in each grid cell, so as to predict the remaining service life of the road structure.

9. A device, characterized in that, The device includes: At least one processor; and A memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform the method for monitoring stress and damage to a road structure as described in any one of claims 1 to 7.

10. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores executable instructions that, when executed, cause the machine to perform the method for monitoring road structure stress and damage as described in any one of claims 1 to 7.

Images